Finbar Quinn, Sofiia Kovalenko
Introduction & concept
Sunlight is typically represented as a trajectory across the sky, reducing a spatial phenomenon to a simple curve. In reality, solar exposure occupies a three-dimensional volume that interacts with architectural space over time.
This project treats sunlight as an active spatial force rather than a passive environmental factor. Instead of drawing the sun’s path, it constructs the space through which light must travel, revealing invisible corridors that define where architecture can and cannot exist. In this way, solar access becomes not a constraint applied after design, but a formative condition that shapes architectural capacity from the outset.
Early Design Sketches
Project Overview
This project generates the maximum buildable volume on a site while ensuring that each existing window receives at least 90 minutes of direct sunlight within a defined time range (e.g., 8:00–16:00).
The workflow operates inversely. Rather than testing a proposed design for compliance, the system identifies volumes that would obstruct required sunlight, subtracts them from the site, and leaves behind only the space that can be legally built.
System Components
The methodology integrates three core elements:
- Environmental Simulation: Sun position analysis and ray-based sunlight verification.
- Voxel-Based Spatial Subtraction: The site is divided into voxels (3D pixels). Voxels that block required sunlight are removed.
- Evolutionary Optimization: Using Galapagos to maximize buildable volume while maintaining solar compliance constraints.
Core Logic
The process operates in three conceptual layers:
- Sunlight Requirement Generation
- Spatial Subtraction via Voxels
- Optimization of Sunlight Allocation using Galapagos
Each layer builds on the previous one.
So, how the algorithm works?
The workflow consists of five main stages.
- Input Parameters
Geometry Inputs
- Window geometries (existing windows requiring sunlight)
- Plot boundary
- Maximum allowable build height
From these, a maximum theoretical buildable envelope is generated.
Resolution Inputs
- Voxel size (controls spatial precision)
- Point cloud density within sunbeams (controls subtraction accuracy)
Higher precision increases computational cost.
Sun Analysis Inputs (LADYBUG PLUGIN)
- Analysis time range (e.g., 8:00–16:00)
- Required direct sunlight duration (default: 90 minutes)
- Time step between segments (default: 15 minutes)
- Sunbeam length
This produces multiple candidate 90-minute sunlight segments for each window.
Geometry input
2. Time Segmentation and Sunbeam Generation
Within the defined time range:
- A new 90-minute segment begins every 15 minutes.
- Each window therefore has multiple possible valid sunlight windows.
For each window and each segment:
- Sun positions are sampled.
- Rays are cast from the window toward the sun.
- Rays are extended to a defined distance.
- A point cloud is generated along these rays.
- This point cloud defines a 3D sunbeam volume.
This volume represents space that must remain empty to preserve that specific 90-minute sunlight condition.
2.1 Construction of viable lines for sunbeams
Generating directional line vectors that define the geometric framework for constructing the sunbeam Brep volume.
2.2 Sunbeam Point Organization
The generated sunbeam lines are organized into a structured data tree, assigning each line to a specific window corner. This ensures clean grouping and enables accurate Brep construction.
2.3 Sunbeam Construction
The organized points are used to construct a closed Brep volume representing the spatial extent of each sunbeam. This volume defines the space that must remain unobstructed.
2.4 Sunbeam Selection
Since multiple 90-minute segments exist, several sunbeam volumes are generated per window. One volume per window is selected using Gene Pool, which acts as a set of decision variables for evolutionary optimization in Galapagos.
3. Voxel Grid Construction
A voxel grid is generated within:
- The plot boundary
- Up to the defined maximum height
Each voxel represents a unit of potentially buildable space.
At this stage, the grid is completely solid — representing the maximum theoretical buildable mass.
3.1 Voxel Center Grid Generation
A grid of voxel center points is generated inside the plot volume, which must be sufficiently large to encompass all potential solar constraint volumes. These points define the spatial framework for voxel creation.
3.2 Voxel Generator
At this stage, a single voxel box is defined as a base volumetric unit. This voxel will later be instantiated at the remaining grid center points after the subtraction and filtering processes.
4. Subtractive Logic
For a selected sunlight segment per window:
- The corresponding sunbeam volume is tested against the voxel grid.
- Any voxel intersecting the sunbeam point cloud is deleted.
- This guarantees that the assigned 90-minute segment remains unobstructed.
After processing all windows, the remaining voxels represent compliant buildable space.
4.1 Deleting Shading Voxels
The sunbeam Breps are populated with point clouds, and the closest points to each voxel center are evaluated. Voxels associated with points inside the sunbeam volume are identified and deleted.
4.2 Pre-Optimization Output
5. Optimization with Galapagos
This is where the system becomes generative rather than deterministic. Each window has multiple valid 90-minute segments. Choosing different segments leads to different spatial deletions. Instead of manually selecting segments, Galapagos is used as an evolutionary solver.
Genome (Variables)
- Each gene represents the selected segment index for one window.
- The full genome defines one complete sunlight allocation configuration.
Fitness Function
- The objective is to maximize remaining voxel count.
- More remaining voxels = more buildable volume.
Galapagos evolves combinations of segment assignments and searches for solutions that:
- Satisfy the 90-minute requirement for each window
- Minimize total removed space
Maximize overall buildable mass
Why Optimization Is Necessary?
Without optimization, the selection of sunlight segments remains arbitrary, causing sunbeam volumes to disperse unpredictably in space. As a result, the buildable volume becomes fragmented and inefficient.
With Galapagos, sunlight corridors begin to spatially align and overlap. Windows tend to share similar solar paths, consolidating void spaces and producing a denser, more continuous buildable mass.
In essence, the system learns to compress environmental constraints into the smallest possible spatial footprint.
Summary
1.Adjustable Parameters Summary
Geometry
- Window geometries
- Plot boundary
- Maximum height
Sun Parameters
- Start time
- End time
- Segment length
- Segment step interval
- Sunbeam length
Resolution
- Voxel size
- Sunbeam sampling density
Optimization
- Galapagos genome (segment index per window)
- Fitness = remaining voxel count
2. Output
The system produces:
- A voxel-based 3D model of the maximum buildable volume
- A spatial visualization of sunlight-protected corridors
- An optimized sunlight allocation strategy
- Quantitative feedback (voxel count as fitness score)
Conceptual Framing
This workflow transforms a regulatory requirement (“Each window must receive 90 minutes of direct sunlight”) into a spatial optimization problem.
Sunlight becomes an active design agent that carves voids from a solid mass, while Galapagos reorganizes those voids to maximize development potential.
Rather than checking compliance after design, the system generates compliant spatial capacity directly from environmental logic.
Download the scripts and Rhino test file below.